Influences of Cleanliness, Carbide Size, Hardness, and Residual Stresses on Long‐Life Fatigue Strength and Crack‐Initiation Mechanisms in Carbide‐Rich Cold‐Work Tool Steels Using Normalized S–N Curves and Database‐Driven Validation of Murakami's √Are
Lennart Mirko Scholl, Frederik Tegeder, Dhirendar Chauhan, Bengt Hallstedt, Tobias Hajeck, Alexander Bezold, Christoph BroeckmannWear‐resistant, carbide‐rich cold‐work tool steels are widely used in massive cold‐forming, where tool life is limited by wear and fatigue. Yet the coupled influences of steelmaking route (cleanliness/carbide structure), heat treatment (hardness), and surface integrity on long‐life fatigue (LLF) strength at cycles remain insufficiently quantified for conventionally manufactured (non‐AM) grades. Rotating‐bending tests () were performed at four to five application‐relevant hardness levels on AISI D2 and AISI A11 (V10) produced by ingot casting + hot working (IC + HW) or gas‐atomized powder metallurgy consolidated by hot isostatic pressing + hot working (PM‐HIP + HW). For AISI A11, process cleanliness was systematically varied, yielding three different microstructural conditions with increasing cleanliness: V10 PM‐L 1 , V10 PM‐L 2 , and V10 PM‐L 3 . Statistical fracture‐surface analysis links fatigue strength to intrinsic defects (carbides, nonmetallic inclusions, and agglomerates) via extreme‐value √area statistics and enables database‐driven assessment of Murakami's √area framework using normalized S–N curves. Fatigue strength increases with hardness only when defect extremes are sufficiently small; strengths 1,000 MPa under rotating‐bending require √area max 60 μm, and further gains require √area max 20 μm. Fabrication‐ and finishing‐induced compressive residual stresses produce surface‐shielding effects that shift normalized run‐out data upward and can attenuate or mask the reverse hardness effect (RHE). Conversely, downward shifts for some fractures are consistent with defect‐size‐controlled RHE behavior that is intentionally not captured by the basic √area model.